Skip to contentSkip to conversation

KVM: nSVM: always intercept VMLOAD/VMSAVE when nested (CVE-2021-3656)

PATCH In review Paolo Bonzini 3 participants 4 messages Aug 16, 2021

Subsystems (2)
lore
Share & tools
lore

Paolo Bonzini

author maintainer
#0 lore raw

PATCH 1/1 KVM: nSVM: always intercept VMLOAD/VMSAVE when nested (CVE-2021-3656)

From: Maxim Levitsky <[email protected]>

[ upstream commit c7dfa4009965a9b2d7b329ee970eb8da0d32f0bc ]

If L1 disables VMLOAD/VMSAVE intercepts, and doesn't enable Virtual VMLOAD/VMSAVE (currently not supported for the nested hypervisor), then VMLOAD/VMSAVE must operate on the L1 physical memory, which is only possible by making L0 intercept these instructions.

Failure to do so allowed the nested guest to run VMLOAD/VMSAVE unintercepted, and thus read/write portions of the host physical memory.

Fixes 89c8a4984fc9 ("KVM: SVM: Enable Virtual VMLOAD VMSAVE feature")

---
The above upstream SHA1 is still on its way to Linus

 arch/x86/kvm/svm.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/svm.c +3 -0
index 0dfd0af61a29..439aecbbdd1e 100644
@@ -389,6 +389,9 @@ static void recalc_intercepts(struct vcpu_svm *svm)
 	c->intercept_dr = h->intercept_dr | g->intercept_dr;
 	c->intercept_exceptions = h->intercept_exceptions | g->intercept_exceptions;
 	c->intercept = h->intercept | g->intercept;
+
+	c->intercept |= (1ULL << INTERCEPT_VMLOAD);
+	c->intercept |= (1ULL << INTERCEPT_VMSAVE);
 }
 
 static inline struct vmcb *get_host_vmcb(struct vcpu_svm *svm)
Signature · 1 line
2.26.3
maintainer
#1 lore raw
Quoted 3 lines · Paolo Bonzini

From: Maxim Levitsky <[email protected]>

[ upstream commit c7dfa4009965a9b2d7b329ee970eb8da0d32f0bc ]

This is not a commit in Linus's tree :(

author maintainer
#2 lore raw
Quoted 4 lines · Greg KH
Quoted 2 lines

[ upstream commit c7dfa4009965a9b2d7b329ee970eb8da0d32f0bc ]

This is not a commit in Linus's tree:(

Not yet, that was remarked further down in the message. I sent the stable patches at the same time as the embargo lifted.

Paolo

#3 lore raw
In reply to Paolo Bonzini
Quoted 12 lines · Paolo Bonzini

From: Maxim Levitsky <[email protected]>

[ upstream commit c7dfa4009965a9b2d7b329ee970eb8da0d32f0bc ]

If L1 disables VMLOAD/VMSAVE intercepts, and doesn't enable Virtual VMLOAD/VMSAVE (currently not supported for the nested hypervisor), then VMLOAD/VMSAVE must operate on the L1 physical memory, which is only possible by making L0 intercept these instructions.

Failure to do so allowed the nested guest to run VMLOAD/VMSAVE unintercepted, and thus read/write portions of the host physical memory.

Quoted 7 lines
Fixes 89c8a4984fc9 ("KVM: SVM: Enable Virtual VMLOAD VMSAVE feature")

---
The above upstream SHA1 is still on its way to Linus

Ah, missed this down here, I read top-to-bottom and stopped at the second line :)

Anyway, will wait for Linus to pick these up first.

thanks,

greg k-h